MPEG-2 video compression

I assume that you are familiar with

Introduction to video compression, on page 147, and with JPEG, M-JPEG, and DV, described in the preceding two chapters.

ISO/IEC 13818-1, Generic coding of moving pictures and associated audio information: Systems [MPEG-2], also published as ITU-T H.220.0.

ISO/IEC 13818-2, Generic coding of moving pictures and associated audio information: Video [MPEG-2], also published as ITU-T H.262.

The DCT-based intrafield or intraframe compression at the heart of M-JPEG is suitable for video production; however, for distribution, dramatically higher compression ratios can be obtained by using interframe coding. MPEG-2 video compression exploits temporal coherence – the statistical likelihood that successive pictures in a video sequence are very similar. MPEG-2’s intended application ranges from below SD to beyond HD; the intended bit rate ranges from about 1.5 Mb/s to well over 20 Mb/s. MPEG-2 also defines audio compression, and provides for the transport of video with associated audio.

MPEG-2 refers to a suite of standards, promulgated jointly by ISO/IEC and ITU-T. The suite starts with Part 1: Systems and Part 2: Video, cited in the margin, which are jointly published by ISO, IEC, and ITU-T. Six other parts are jointly published by ISO and IEC – Part 3: Audio; Part 4: Conformance testing; Part 5: Software simulation; Part 6: Extensions for DSM-CC; Part 7: Advanced Audio Coding (AAC); Part 9: Extension for real time interface for systems decoders; and Part 10: Conformance extensions for Digital Storage Media Command and Control (DSM-CC). The projected Part 8, for 10-bit video, was discontinued. MPEG-2 standards were first issued in 1996; subsequently, several corrigenda and amendments have been issued.

MPEG-2 specifies exactly what constitutes a legal bitstream: A legal (“conformant”) encoder generates only legal bitstreams; a legal decoder correctly decodes any legal bitstream. MPEG-2 does not standardize how an encoder accomplishes compression!

MPEG-2 specifies several algorithmic features – such as arbitrary frame rate, and 4:4:4 chroma subsampling – that are not permitted in any standard profile. These features are unlikely to see commercialization.

The MPEG-2 standard implicitly defines exactly how a decoder reconstructs pictures data from a coded bitstream, without dictating the implementation of the decoder. MPEG-2 explicitly avoids specifying what it calls the “display process” – how reconstructed pictures are displayed. Most MPEG-2 decoder implementations have flexible output formats; however, MPEG-2 decoder equipment is ordinarily designed to output

a specific raster standard.

An MPEG-2 bitstream may represent interlaced or progressive pictures. Typical decoder equipment outputs either interlace or progressive signals. Certain decoder equipment has the capability to switch between the two output formats. Because interlaced scanning remains dominant in consumer electronics – both in SD and in HD – a decoder system must be capable of producing an interlaced signal from a progressive sequence. Also, it is a practical necessity for an MPEG-2 decoder to have spatial resampling capability: If an HD MPEG-2 decoder is presented with an SD sequence, consumers would complain if reconstructed pictures were not upconverted for display in HD.

MPEG-2 profiles and levels

An MPEG-2 bitstream can potentially invoke many algorithmic features – some practitioners call them “tools” – at a decoder. Also, a bitstream can reflect many possible parameter values. The MPEG-2 standard classifies bitstreams and decoders in a matrix of profiles and levels.

Profiles constrain the algorithmic features potentially used by an encoder, present in a bitstream, or implemented in a decoder. The higher the profile, the more complexity is required of the decoder. MPEG-2 defines six profiles: Simple (SP), Main (MP), 4:2:2 (422P), SNR,

Spatial (Spt), High (HP), and Multiview (MVP).

Levels place restrictions on parameter values used by an encoder or decoder. The higher the level, the more memory or data throughput is required of a decoder.

MPEG-2 defines four levels: Low (LL), Main (ML), High-1440 (H14), and High (HL).

A profile and level combination is indicated by profile and level separated by an at sign – for example, MP@ML or MP@HL. The SNR, Spatial, High, and

422P@ML allows 608 lines at

25 Hz frame rate, but is limited to 512 lines at 29.97 and 30 Hz frame rates.

SMPTE 308M, Television – MPEG-2

4:2:2 Profile at High Level.

The simple profile has no B-pictures. Prohibition of B-pictures minimizes encoding latency, and minimizes buffer storage at the decoder. However, the simple profile lacks the compression efficiency of B-pictures.

Of the eight combinations in Table 47.1, only two are commercially important to television. MP@ML is used for SD distribution, and for DVD, at rates from about 2 Mb/s to about 6 Mb/s. MP@HL is used for HD distribution, usually between 10 Mb/s and 20 Mb/s.

The 4:2:2 profile allows 4:2:2 chroma subsampling; it is intended for use in television production. The major reason for a separate 4:2:2 profile is that main profile disallows 4:2:2 chroma subsampling. MPEG-2’s high profile allows 4:2:2 subsampling, but to require highprofile conformance would oblige a decoder to handle SNR and spatial scalability. 422P@ML is used in the studio, as Sony MPEG IMX, at bit rates between

30 Mb/s and 50 Mb/s. Some numerical parameter limits of main and 4:2:2 profiles are presented in Table 47.2 above.

MPEG-2 defines 4:2:2 profile at high level (422P@HL). In addition to MPEG-2’s requirements for 422P@HL, SMPTE 308M imposes these restrictions on